dissenting.
For more than twenty years, the State of Ohio has displayed a remarkable inability to competently prosecute Joe D’Ambrosio. In 1988, the state botched D’Ambrosio’s trial by violating Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), an error that later came back to bite it in federal habeas review. D’Ambrosio v. Bagley, 527 F.3d 489 (6th Cir.2008). Then, after D’Ambrosio was issued a conditional writ of habeas corpus, the state misunderstood the district court’s clear, binary order and attempted to simultaneously comply with both of the two alternatives it was given: retry D’Ambrosio within 180 days or vacate his conviction and release him. After failing to complete only the prompt-retrial option, the hapless state marched back to district court and, making D’Ambrosio’s argument for him, conceded that it failed to comply with the conditional writ. Ever since, the state has asserted a variety of confused jurisdictional arguments in both the district court and *391in this court, and it is now in the position where it may have to let a man it believes to be a murderer go forever free. Whether D’Ambrosio deserves that windfall I cannot say, although, after more than twenty years of bungling his criminal proceedings, surely the state deserves that penalty. But this is a case about jurisdiction. Whatever the equities in D’Ambrosio’s favor, the state of Ohio — by sheer luck and nothing more — managed to do just enough to strip the district court of its power to decide the issue. Accordingly, I would reverse the district court and dismiss the case.
I
Because the state of Ohio vacated D’Ambrosio’s unconstitutional conviction and released him from any confinement pursuant to that conviction, thereby complying with the district court’s conditional writ of habeas corpus, the district court lacked jurisdiction to issue an unconditional writ. In my view, the majority commits two major errors in coming to an opposite conclusion.
First, the majority erroneously considers compliance with the precise terms of the conditional-writ order to be the primary jurisdictional inquiry, and in doing so loses the forest for the trees. It is worth reviewing what exactly a conditional writ is. When a federal court holds that a prisoner’s sentence is unconstitutional, the prisoner is instantly entitled to an unconditional writ of habeas corpus — immediate release and vacation of the conviction. However, as a convenience to the state, the federal court may instead issue a conditional writ, which delays the issuance of the unconditional writ so that the state has an opportunity to cure its own constitutional violation. Wilkinson v. Dotson, 544 U.S. 74, 86-87, 125 S.Ct. 1242, 161 L.Ed.2d 253 (2005) (Scalia, J., concurring); Hilton v. Braunskill, 481 U.S. 770, 775, 107 S.Ct. 2113, 95 L.Ed.2d 724 (1987); Gentry v. Deuth, 456 F.3d 687, 692 (6th Cir.2006). The backdrop of a conditional writ, then, is the release and vacation provided by an unconditional writ, which fully vindicates the prisoner’s constitutional rights, and the sole purpose of a conditional writ is to afford the state a period of time to accomplish that result on its own before the district court must directly interfere with the state’s judgment. Federal courts’ preference for conditional writs is obvious in light of the concepts of federalism embodied in the structure of federal habeas corpus. See 28 U.S.C. § 2254; Coleman v. Thompson, 501 U.S. 722, 726, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991) (“This is a case about federalism. It concerns the respect that federal courts owe the States ... when reviewing the claims of state prisoners in federal habeas corpus.”); see generally Younger v. Harris, 401 U.S. 37, 44-45, 91 S.Ct. 746, 27 L.Ed.2d 669 (1971) (“ ‘Our Federalism’ ... [is] a system in which ... the National Government, anxious though it may be to vindicate and protect federal rights and federal interests, always endeavors to do so in ways that will not unduly interfere with the legitimate activities of the States.”).
With that in mind, the fundamental question for a district court when considering whether the conditional writ has been complied with is whether the state has fully vindicated the prisoner’s rights, thereby rendering an unconditional writ unnecessary. Indeed, what an unconditional writ accomplishes — release from unconstitutional custody — is precisely what a conditional writ encourages the state to accomplish on its own, and if a state accomplishes this goal, the mission of the district court is at an end. A district court could not, for example, issue a conditional writ that required the state both to vacate the conviction and to conduct a new trial within a certain period of time. By subjecting the state to more federal intrusion than would an unconditional writ, such a *392conditional writ would not be an “accommodation[ ] accorded to the state.” Satterlee v. Wolfenbarger, 453 F.3d 362, 369 (6th Cir.2006). More significantly, such an order would do more than vindicate the prisoner’s constitutional rights and, because it would usurp from the state courts the power to ensure, in the first instance, that a new state trial complies with constitutional standards, it would also run squarely into the exhaustion problems addressed by Pitchess v. Davis, 421 U.S. 482, 95 S.Ct. 1748, 44 L.Ed.2d 317 (1975), which held that a district court lacks authority to supervise a retrial conducted pursuant to a conditional writ of habeas corpus. Id. at 490, 95 S.Ct. 1748.
Because a vacation of an unconstitutional conviction fully vindicates a prisoner’s rights, it necessarily complies with a conditional writ — even if the writ does not explicitly provide vacation as an option. Pickens v. Howes, 549 F.3d 377, 379, 384 (6th Cir.2008) (holding that, where conditional writ required the state to allow the prisoner to withdraw his guilty plea, the prisoner’s constitutional rights “were fully vindicated by the state court’s decision vacating his ... conviction.”). As a result, although there is no question that a district court retains jurisdiction to determine whether the conditional writ has been complied with, Gentry v. Deuth, 456 F.3d at 692, the scope of that inquiry is limited by the threshold question of whether there exists an unconstitutional conviction. Eddleman follows from this straightforward principle, holding that, where the state has vacated the unconstitutional conviction, it has ipso facto complied with the conditional writ and the district court’s jurisdiction is at an end. See Eddleman v. McKee, 586 F.3d 409, 413 (6th Cir.2009). Accordingly, without first considering whether D’Ambrosio’s conviction had been vacated, Pitchess and Fisher cannot be distin: guished on the grounds that, in those cases, the state complied with the conditional writ.
Second, when later considering whether the state had vacated D’Ambrosio’s conviction, the majority erroneously focuses on the state’s confused arguments and actions before the district court. To be sure, the state conceded through its actions that the district court had jurisdiction to issue the unconditional writ. But that is neither here nor there. Independent of the state’s concession, the district court had “an independent obligation to investigate and police the boundaries of [its] own jurisdiction,” and “no action of the parties can confer subject-matter jurisdiction upon a federal court” where it is lacking. In re Cassim, 594 F.3d 432, 439 (6th Cir.2010) (quoting Douglas v. E.G. Baldwin & Assocs., Inc., 150 F.3d 604, 607 (6th Cir.1998)); Ins. Corp. of Ireland v. Compagnie des Bauxites de Guinee, 456 U.S. 694, 702, 102 S.Ct. 2099, 72 L.Ed.2d 492 (1982). The prosecution’s actions and beliefs do not settle the question, and that conclusion is punctuated in this case, where the prosecution’s understanding of the legal landscape has been consistently faulty.
Notwithstanding the state’s confused representations to the district court and to this court, it is clear that D’Ambrosio’s conviction had been vacated prior to the issuance of the unconditional writ. Contrary to the majority’s suggestion that the state-court judge merely used the term “vacate” on occasion and to no real effect, the record manifests that both the state court and the parties involved were in agreement that the unconstitutional conviction had been vacated and treated it as such, thereby placing this case squarely within the holding of Eddleman, 586 F.3d at 412.
On October 23, 2008, D’Ambrosio was transferred from the state penitentiary in Youngstown to Cuyahoga County Jail, *393pending the result of the retrial. On February 17, 2009, D’Ambrosio filed a motion in state court to set bail, stating: “Defendant is being held in jail without bail, pending trial of this case. At his arraignment on October 14, 1988, bail was set at $200,000, cash. He was convicted and sentenced, but the conviction was later vacated.” In its brief in opposition to D’Ambrosio’s motion, filed on February 20, 2009, the prosecution conceded that D’Ambrosio was no longer being held pursuant to an existing conviction and sentence. Although the prosecution did not use the term “vacated” in its brief, it agreed that the conviction had been “overturned” and no longer had any effect.
More to the point, the state court was also in agreement that the conviction had been vacated. In its February 24, 2009, order granting bail, the state court twice noted that D’Ambrosio’s conviction had been vacated, not in mere passing, but as the primary justification of its decision on the matter. The state court rejected the prosecution’s argument that the presumption of guilt was great in light of his prior conviction solely because “[t]hat conviction has since been vacated.” In deciding to grant bail, the state court also explicitly relied on the fact that D’Ambrosio had no prior felony record: “[t]he defendant stands before the court today as he did at his October 14, 1988 arraignment, with no prior felony record.”
In my view, as of February 24, 2009, all the parties were in agreement that D’Ambrosio’s conviction had been vacated, and they were correct — the record reflects that the prior conviction had no continuing effect by this time. Indeed, shortly after the state court’s order, D’Ambrosio made bail and was released, pending the outcome of his retrial- — activity clearly inconsistent with a conclusion that he remained in custody pursuant to an unconstitutional conviction, as opposed to a new indictment. Therefore, because D’Ambrosio’s prior conviction was no more, the district court lacked jurisdiction to issue the unconditional writ.
II
Because I would hold that the district court lacked jurisdiction to issue the unconditional writ, as a matter of course, I would hold that it lacked jurisdiction to revisit its order, pursuant to a Rule 60(b) motion, and change its mind as to whether reprosecution should be barred. See Charter Tp. of Muskegon v. City of Muskegon, 303 F.3d 755, 762 (6th Cir.2002) (“A Rule 60(b) motion is considered a continuation of the original proceeding. If the district court had jurisdiction when the suit was filed, it has jurisdiction to entertain a Rule 60(b) motion.”) (internal quotation omitted). However, I would reach the same outcome, albeit on non-jurisdictional grounds, even if I agreed that the district court had jurisdiction to issue the unconditional writ.
Pitchess makes clear that Rule 60(b) cannot be used to circumvent section 2254’s exhaustion requirement, and its holding directly controls this case. Although exhaustion is non-jurisdictional in nature, White v. Mitchell, 431 F.3d 517, 526 (6th Cir.2005), it similarly bars access to federal courts unless expressly waived by the state, 28 U.S.C. § 2254(b)(3), and there can be no question that the state did not so waive it here.1 In Pitchess, the *394district court granted the petitioner a conditional writ that ordered the state to commence a retrial, and the state began retrial proceedings, in full compliance with the conditional writ. 421 U.S. at 483-84, 95 S.Ct. 1748. During this time, the petitioner discovered that crucial evidence had been destroyed prior to the issuance of the conditional writ. Id. at 484, 95 S.Ct. 1748. The petitioner then returned to federal court and requested, pursuant to Rule 60(b), an order barring reprosecution. Id. at 485, 95 S.Ct. 1748. The district court granted the order and the Ninth Circuit affirmed, reasoning that, because the district court would have granted the requested relief if it had known about the destroyed evidence at the time it issued the conditional writ, Rule 60(b) justified the grant of relief after the fact. Id. at 485-86, 489, 95 S.Ct. 1748. That justification was flatly rejected by the Supreme Court, which held that Rule 60(b) does not alter the statutory exhaustion requirement that applies to habeas proceedings:
Respondent failed to exhaust available state remedies on the claim which formed the basis for the unconditional writ, and he is entitled to no relief based upon a claim with respect to which state remedies have not been exhausted. Neither Rule 60(b), 28 U.S.C. § 2254, nor the two read together, permit a federal habeas court to maintain a continuing supervision over a retrial conducted pursuant to a conditional writ granted by the habeas court.
Id. at 490, 95 S.Ct. 1748. The key language — “[njeither Rule 60(b), 28 U.S.C. § 2254, nor the two read together” — is in reality just as applicable to this case as it sounds, and the majority’s attempts to distinguish Pitchess simply fall flat.
First, the majority argues that, in Pitchess, the state had complied with the conditional writ, whereas here, the State of Ohio did not. Even assuming that statement were true, it makes no difference. To be sure, the state’s purported failure to comply with the conditional writ does impact the question of whether the district court had jurisdiction to issue the unconditional order — if the conviction remains in effect, then the district court retains jurisdiction. However, any failure to comply with the initial conditional writ has absolutely no bearing on the question of the district court’s authority to issue a new order after the unconditional writ has issued. Indeed, it is the fact that the state in Pitchess complied with the conditional writ that makes it on point with this case. In Pitchess, the state’s compliance with the conditional order is what vindicated the prisoner’s rights — as though an unconditional writ has issued — and ended the district court’s continued authority over the matter. Here, the unconditional writ, in fact, issued. There is absolutely no question that, at the time D’Ambrosio filed his Rule 60(b) motion, his constitutional rights had been fully vindicated. Accordingly, at the time the motions were filed, D’Ambrosio was in the exact same position as the petitioner in Pitchess. Any issues with the earlier conditional writ are simply irrelevant.
Second, the majority argues that, because D’Ambrosio based his argument for barring reprosecution on the Brady claim that formed the basis of the conditional writ and not on events that occurred after the district court granted the unconditional writ, this case “does not present the same sort of exhaustion issues” as Pitchess. That attempted distinction is both factually and legally incorrect. A reading of the Pitchess opinion makes clear, just as is the case here, the requested relief was based on events that occurred before the district court’s original order. Indeed, the majority’s prior-events argument was the precise justification put forth by the Ninth Circuit — and resoundingly rejected by the Supreme Court. Next, D’Ambrosio certainly *395did not premise his argument for barring reprosecution on the same Brady claim that had resulted in the earlier habeas relief. In his Rule 60(b) motion, D’Ambrosio’s argument was not that the state withheld exculpatory information in his 1988 trial, but rather that, without the ability to cross-examine the state’s dead witness, he could not mount an adequate defense at retrial. That is not the old Brady claim. It is an entirely new substantial-delay claim. See United States v. Marion, 404 U.S. 307, 324, 92 S.Ct. 455, 30 L.Ed.2d 468 (1971) (noting that “the Due Process Clause of the Fifth Amendment would require dismissal of the indictment if it were shown at trial that the pre-indictment delay in this case caused substantial prejudice to [the defendant’s] rights to a fair trial,” and distinguishing such a claim from Brady). D’Ambrosio did not present this claim to the state courts and, as a result, the district court was without authority to consider it. Pitchess, 421 U.S. at 489-90, 95 S.Ct. 1748.
D’Ambrosio made the argument — which the majority appears to buy in full — that his new claim is actually the old claim because there is a causal connection between the two. True, D’Ambrosio’s new claim was in some sense caused by the state’s 1988 violation of Brady: had the state done things correctly the first time around, the witness would have been alive and available for an effective cross examination (assuming no other judicial interruption with a concomitant delay). Of course, that is the case every time there is any problem with a retrial, yet federal courts may not review even such connected issues until the new judgment is final and such issues are raised in a new habeas petition following an unsuccessful resort to all available state remedies. See Pitchess, 421 U.S. at 486-87, 95 S.Ct. 1748; 28 U.S.C. §§ 2241, 2254. In the same sense, the new claim was affected by the state’s failure to commence a trial within 180 days: had it done so, the witness might not have expired before the retrial. But that speaks only to, perhaps, the state’s culpability; it does not somehow convert D’Ambrosio’s unexhausted, new claim— based upon different facts and legal standards — into his old Brady claim. Further, that sort of causation was equally at issue in Eddleman, where the petitioner’s Rule 60(b) motion argued that he was prejudiced by the state’s delay in conducting his retrial, in violation of the conditional writ. Although the causation argument D’Ambrosio makes here was equally applicable in Eddleman, this court held that “the district court proceeded effectively to adjudicate a speedy-trial claim that had never been presented to, much less ruled upon by, the Michigan state courts. With due respect, the district court acted out of turn in doing so.” 586 F.3d at 413. That holding applies with full force here.
The majority’s reliance on Satterlee and Girts is misguided. Notably, because the Supreme Court’s decision in Pitchess is precisely on point, even if those cases stand for the proposition the majority credits them with, they are wrongly decided to the extent they conflict with Pitchess. But that is a non-issue, because, contrary to the majority’s assertion, Satterlee and Girts neither hold nor suggest that a district court may issue new orders or materially amend old ones once an unconditional writ has been granted and complied with. In Satterlee, the district court granted an unconditional writ that released the prisoner from custody. Satterlee, 453 F.3d at 365. The state appealed, and this court affirmed the grant of the unconditional writ, remanding to the district court to clarify whether its pre-release order was intended to allow the state to reprosecute, which was not made clear in the court’s initial order. 453 F.3d at 370 (“[W]e instruct the district court to clarify on remand which of these meanings *396it intended, and if it meant to forbid re-prosecution, to justify its conclusion.”) (emphasis supplied). The Satterlee court, then, did not suggest that a district court may issue a new order after the conviction has been vacated, but only that it can clear up an earlier, vague decision. And in Girts, this court similarly held that the district court could clarify its order granting an unconditional writ, where two days later and hours after the release of the petitioner, the court explained that it did not intend to bar reprosecution in that initial order, and that explanation was not contrary to the text of the initial order. Girts v. Yanai, 600 F.3d 576, 581 (6th Cir.2010). Here, on the other hand, the district court first issued an order that explicitly declined to bar reprosecution and then later, after the conviction had been vacated, issued a new order barring re-prosecution. This is fundamentally at odds with Girts, which explicitly recognized the applicability of Pitchess and made the narrow scope of its holding abundantly clear:
We do not hold that a district court can materially amend its judgment after the state has released a petitioner, and we acknbwledge binding Supreme Court authority that a habeas court does not maintain “continuing supervision over a retrial conducted pursuant to a conditional writ granted by the habeas court.” Pitchess v. Davis, 421 U.S. 482, 490, 95 S.Ct. 1748, 44 L.Ed.2d 317 (1975). In the unique factual setting of this case, we simply read the November 5, 2008 order as clarifying the November 3, 2008 order, in which the district court granted an unconditional writ and in which the district court was free to decide whether to bar retrial.
600 F.3d at 582. Neither Girts nor Satterlee, then, can be read to allow a district court to issue new orders after the underlying conviction has been vacated, but rather that the district court retains only the power to elucidate an earlier, vague decision. The Girts court noted that the limits of this clarification power must fall short of “materially amending]” an earlier judgment, which is precisely what the district court did here. Accordingly, even if I agreed that the district court retained jurisdiction to grant the unconditional writ, I would reverse its later order barring re-prosecution pursuant to Pitchess — that order blatantly exercises exactly the type of “continuing supervision” barred by Pitchess and contemplated by Girts.
Ill
The majority correctly notes that the warden does not challenge the district court’s order as an abuse of discretion. That, in my view, was bad strategy, albeit consistent with the state’s body of conduct over the past twenty-odd years. In any case, the issue was neither briefed nor otherwise raised by either party, and that is sufficient to end the inquiry. I note that it is far from apparent to me that the death of the prosecution’s star witness before D’Ambrosio’s retrial prejudiced him to such an extent that the extreme measure of barring the retrial altogether was justified. At first blush, the witness’s death appears to be beneficial to D’Ambrosio, particularly in light of the state court’s decision not to allow the witness’s prior testimony to be presented at the retrial. Further, even if D’Ambrosio were to be greatly prejudiced at retrial by the state’s misconduct, he could bring that due-process claim to the state courts on direct review. Because any potential prejudice appears to be curable through the ordinary process, the district court’s decision to bar reprosecution is, if not an abuse of discretion, at least highly suspect. Scott v. Bock, 576 F.Supp.2d 832, 837-38 (E.D.Mich.2008) (“As the term ‘extraordinary’ applies, courts should ... bar[ ] re-prosecution sparingly, reserving this dras*397tic remedy for the most outrageous cases.... Unless the constitutional violation is not redressable at a new trial, or the second trial would be especially unjust, the bar on reprosecution is difficult to justify”) (citations omitted).
IV
In my view, the majority’s opinion is unmoored from both the facts of this case and the applicable case law of both this circuit and the Supreme Court. Moreover, I find particularly troubling that the majority opinion permits federal courts to order habeas relief not only based on the conditions leading to an underlying conviction, but also based on the conditions of an ongoing state criminal proceeding. Because habeas relief may be granted only from final criminal judgments, and only under narrow circumstances, see 28 U.S.C. §§ 2241, 2254, I remain unconvinced that federal courts, under the cloak of habeas jurisdiction, have the authority to interfere with a state’s ongoing prosecutorial efforts, and that is precisely what the district court did here. I therefore respectfully dissent.
. In its brief, the state heavily relies on Pitchess and its holding regarding exhaustion. Appellant's Br. at 18-20. That the state appeared to mistakenly conclude that Pitchess was about jurisdiction is not an express waiver of the exhaustion arguments it included in its brief, and, in any case, this court may consider the issue sua sponte. See Granberry v. Greer, 481 U.S. 129, 133-34, 107 S.Ct. 1671, 95 L.Ed.2d 119 (1987).